Bedovyy
Bedovyy
@LurkingOnGithub @Jay200001 I have fixed the issue. Please check it works properly.
I added `prefer_brownian` and `deliberate_euler_ancestral_bug` when sampler is `euler_a` and scheduler is not `native`. Please check you can generate same image.
I close the issue. If you have problem, please open another issue.
Thanks for the big thing! I have tested SDXL(Euler a, 28steps) 832x1216 and 1216x1856, and WAN2.1 I2V 480P(from ComfyUI Example but fp8). | | 1x4090 | 1x3090 | 2x3090 |...
> > Thanks for the big thing! > > I have tested SDXL 832x1216 and 1216x1856, and WAN2.1 I2V 480P(from ComfyUI Example but fp8). > > 1x4090 1x3090 2x3090 >...
okay I read the code and the author's article, Only 2 GPUs are working because it has conditioning-based work units. diffusion models usually have 2 conditionings (positive and negative), so...
confirmed working basic SDXL T2I on 2x Arc B580, using torch2.6.1 + intel extension 2.6.10. | | it/s | time taken | |---|---|---| | 1xB580 | 3.68 | 8.92...
I will try.
Thank you for the request! the difficult part of NAI weight to SD weight is that multiple brackets doesn't end at same time. for example, the below prompt is also...
Did you use Image2Image, Inpaint or Vibe? the error tell us request(information for generation + image for I2I or inpaint or something) size is too large. As far as I...